fix(tests): all() on empty document list is vacuously True, masking empty-result regression in list_documents #269

Summary

tests/test_documents_endpoint.py asserts all(doc['doc_id'] == 'doc_abc123' for doc in body['data']) to verify document listing. When body['data'] is an empty list [], all() returns True vacuously — a bug returning an empty document list passes the assertion silently.

Root Cause

tests/test_documents_endpoint.py around line 137:

body = response.json()
assert all(doc['doc_id'] == 'doc_abc123' for doc in body['data'])
# ↑ True when body['data'] == [] because all() on empty iterable is True

There is no assertion on the length of body['data'], so an empty response and a correct response are indistinguishable.

Failure Scenario

1. A regression in list_documents causes it to return {"data": [], "pagination": {...}} for a tenant that has documents.
2. all(... for doc in []) → True.
3. Test passes. The regression ships to production.
4. All users see empty document lists with no error.

Proposed Fix

body = response.json()
docs = body['data']
assert len(docs) == 1, f"Expected 1 document, got {len(docs)}: {docs}"
assert docs[0]['doc_id'] == 'doc_abc123'

Also add a test for page > total_pages returning 422:

def test_list_documents_page_out_of_range_returns_422(client, auth_headers):
    # Upload 1 document, then request page=999
    response = client.get('/v1/documents?page=999', headers=auth_headers)
    assert response.status_code == 422